Are we pondering "calculation" in the sense of a theoretical assessment, or in terms of data analysis?
In case of the former, Artur's answer is exdcellent, Bruno's original treatment is restricted to so-called "strong ferromagnets", though. Van der Laan has worked out a generalization to the more general case of partially filled minority AND majority bands.
Gerrit van der Laan 1998 J. Phys.: Condens. Matter 10 3239 doi:10.1088/0953-8984/10/14/012
In the second case of my previous post, an answer cannot probably be given without knowing what kind of datasets you have or intend to measure.
In general, it takes single crystalline samples to determine anisotropy, and one way to do so is to measure M(H) datasets with the field oriented along different (principal) symmetry directions. You will then have to test and apply different models for how magnetization reversal occurs and fit your data to these models to obtain the anisotropy constants.
in case your system allows for it, measurements of ferromagnetic resonance (FMR) provide a much more elegant and extremely sensitive probe of magnetic anisotropy. Yet again, the interpretation is again a matter of how you interpret what you get.
from my opinion this is a great and complex theory. You must use the surface anisotropy. Please look to the complex description in the Landau Lifschitz Theory for thin layers.